Why We Care about Estimates

Why all this emphasis on accurate estimates?

Because they have a powerful influence on our team's journey. If they're too low, we'll have to drive like maniacs to get to the finish line, risking spinning out of control and crashing. If they're too high, it's not such a bad problem we might end up finishing early but at the wrong place. This beats crashing, and we can always have the customer pick more stories for a short additional drive. But if we get the estimates right, we'll get to the right place at the right time and have a great trip along the way.

Do the estimates have to be perfect? No. The reason we're emphasizing estimates is that over and over we've seen teams underestimate the time needed for testing. In fact, they often leave testing time out of story estimates altogether.

Here's a common scenario we've witnessed: Team X uses up all the time they'd estimated for stories without completing any of the testing tasks. Maybe they even dropped a story, but they're still short of time. Only a day or two is left in the iteration, so they panic and run the acceptance tests manually. Now iteration 2 starts. Not only do they have all the testing tasks for iteration 2, they still have to automate the acceptance tests for iteration 1. The problem snowballs into an avalanche.

If our story estimates include enough time for all the testing tasks, our team won't take on more work than they can actually complete in an iteration. Remember, no story is finished until the acceptance tests have run. That means time for the automation too.

Experienced XP teams can take estimation in stride. They have so much experience, they have a good feel for the effort each story will need. They might be high on some estimates and low on others, but everything will average out, and they can adjust as they go along. If you're working on a team that doesn't have a track record with XP projects, we think you need to pay more attention to the estimates.